Integrated Intrusion Detection Services for. z/os Communications Server. SHARE Session 8329

Transcription

1 Software Group Enterprise Networking Solutions Integrated Intrusion Detection Services for z/os Communications Server SHARE Session 8329 Lin Overby - overbylh@us.ibm.com Lin Overby - overbylh@us.ibm.com March 2, 2011 z/os Communications Server 2011 IBM Corporation Page 1

2 Trademarks and notices The following terms are trademarks or registered trademarks of International Business Machines Corporation in the United States or other countries or both: Advanced Peer-to-Peer Networking AIX alphaworks AnyNet AS/400 BladeCenter Candle CICS DB2 Connect DB2 DRDA e-business on demand e-business (logo) e business(logo) ESCON FICON GDDM HiperSockets HPR Channel Connectivity HyperSwap i5/os (logo) i5/os IBM (logo) IBM IMS IP PrintWay IPDS iseries LANDP Language Environment MQSeries MVS NetView OMEGAMON Open Power OpenPower Operating System/2 Operating System/400 OS/2 OS/390 OS/400 Parallel Sysplex PR/SM pseries RACF Rational Suite Rational Redbooks Redbooks (logo) Sysplex Timer System i5 System p5 System x System z System z9 Tivoli (logo) Tivoli VTAM WebSphere xseries z9 zseries z/architecture z/os z/vm z/vse Java and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United States, other countries, or both. Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both. Intel, Intel Inside (logos), MMX and Pentium are trademarks of Intel Corporation in the United States, other countries, or both. UNIX is a registered trademark of The Open Group in the United States and other countries. Linux is a trademark of Linus Torvalds in the United States, other countries, or both. Red Hat is a trademark of Red Hat, Inc. SUSE LINUX Professional 9.2 from Novell Other company, product, or service names may be trademarks or service marks of others. This information is for planning purposes only. The information herein is subject to change before the products described become generally available. All statements regarding IBM future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only. All performance data contained in this publication was obtained in the specific operating environment and under the conditions described and is presented as an illustration. Performance obtained in other operating environments may vary and customers should conduct their own testing. Refer to for further legal information. Page 2

3 Integrated Intrusion Detection Services z/os Communications Server provides an integrated Intrusion Detection Services (IDS) for TCP/IP. This session will describe the Communications Server IDS and how it can be used to detect intrusion attempts against z/os. This session will cover the following topics IDS Overview Intrusion events detected by z/os IDS IDS Actions Recording Actions Defensive Actions IDS Reports Automation for IDS Working with IDS policy Page 3

4 The Intrusion Threat What is an intrusion? Information Gathering Network and system topology Data location and contents Eavesdropping/Impersonation/Theft On the network/on the server Base for further attacks on others Amplifiers Robot or zombie Denial of Service Attack on availability Intrusions can occur from Internet or intranet End User Attacker Single Packet attacks - exploits system or application vulnerability Multi-Packet attacks - floods systems to exclude useful work Attacks can occur from Internet or intranet Firewalls can provide some level of protection from Internet Perimeter Security Strategy alone may not be sufficient. Considerations: Access permitted from Internet Trust of intranet Attack can be deliberate with malicious intent, or they can result from various forms of errors on nodes in the network Internet Server Enterprise Network or Intranet Firewall Attacker Zombie Attackers Page 4

5 Integrated vs. External Intrusion Detection Concepts Integrated IDS sensor on server Service policy IDS policy IDS policy IDS adapters, IDS agents, event notification Applications Middleware Networking External IDS sensor in network IDS probes and services IDS event data Data is read for processing at server Intrusions can be checked both before and after decryption IPSec IDS checking is done in context Data endpoint, connection state known, statistics available IDS manager, such as Tivoli Security Information and Event Manager Service policy IDS signature files IDS adapters, IDS agents, event notification IDS signatures IDS data scanner IDS event data Data is read by a box in-the-middle in "promiscuous" mode If data is encrypted, IDS scanner cannot detect most signature-based intrusions Integrated IDS on z/os complements external IDS technologies Page 5

6 Intrusion Detection Services Overview IDS Policy administration Install IDS policies into TCP/IP stack Policy Agent IDS policy TRMD SyslogD TCP/UDP/RAW Intrusion event notification IP/ICMP Attack probes Data Link Log event and statistics Dynamic trace of suspicious activity Attack!!! trmdstat utility Trace Automation based on MVS console messages Detail and summary reports Event messages to MVS console z/os in-context IDS broadens overall intrusion detection coverage: Ability to evaluate inbound encrypted data - IDS applied after decryption on the target system Avoids overhead of per packet evaluation against table of known attacks - IDS policy checked after attack detected Detects statistical anomalies real-time - target system has stateful data / internal threshholds that are generally unavailable to external IDSs Policy can control prevention methods on the target, such as connection limiting and packet discard z/os NetView Tivoli Security Information and Event Manager (TSIEM) Events detected Scans Attacks Against Stack Flooding (both TCP and UDP) Defensive methods Packet discard Limit connections Reporting Logging, Event messages to local console, IDS packet trace Notifications to Tivoli NetView and Tivoli Security Information and Event Manager (TSIEM) IDS Policy Samples provided with Configuration Assistant for z/os Communications Server Integrated Intrusion Detection Services under policy control to identify, alert, and document suspicious activity Page 6

7 New Support Planned for z/os V1R13 Extend existing support to IPv6 New attack types: Data hiding TCP Queue Size Global TCP Stall Enterprise Extender protections Page 7

8 IDS Configuration IDS is configured with IDS policy IDS policy defines intrusion events to monitor and actions to take Policy definitions are stored in policy repository File or data set LDAP (no longer being enhanced) Policy Agent reads policy definitions from policy repository Policy definitions are processed by Policy Agent and installed in the TCP/IP stack Page 8

9 Policy Model Overview Basic Policy Objects Policy Condition Policy Rule Policy Action Policy objects relationship: IF condition THEN action Policies consist of several related objects Policy Rule is main object and refers to: Policy Condition Defines IDS conditions which must be met to execute the Policy action Policy Action Defines IDS actions to be performed when Policy Condition is met Page 9

10 z/os Communications Server Security Intrusion Events Types Detected SCAN ATTACK TRAFFIC REGULATION Page 10

11 Intrusion Event Types Supported Scan detection and reporting Intent of scanning is to map the target of the attack Subnet structure, addresses, masks, addresses in-use, system type, op-sys, application ports available, release levels Attack detection, reporting, and prevention Intent is to crash or hang the system Single or multiple packet Traffic regulation for TCP connections and UDP receive queues Could be intended to flood system OR could be an unexpected peak in valid requests Page 11

12 Scanning... the prelude to the attack z/os IDS definition of a scanner Source host that accesses multiple unique resources (ports or interfaces) over a specified time period Installation can specify via policy number of unique events (Threshold) and scan time period (Interval) Categories of scan detection supported Fast scan Many resources rapidly accessed in a short time period (less than 5 minutes) usually less than five minutes, program driven Slow scans Different resources intermittantly accessed over a longer time period (many hours) scanner trying to avoid detection Scan events types supported ICMP scans TCP port scans UDP port scans Page 12

13 Scan Policy Overview Scan policy provides the ability to: Obtain notification and documentation of scanning activity Notify the installation of a detected scan via console message or syslogd message Trace potential scan packets Control the parameters that define a scan: The time interval The threshold number of scan events Reduce level of false positives Exclude well known "legitimate scanners" via exclusion list e.g. network management Specify a scan sensitivity level by port for UDP and TCP highest priority rule for ICMP Page 13

14 Scan Event Counting and Scan Sensitivity Each scan event is internally classified as normal, suspicious or very suspicious Socket state, ICMP type affect this classification Scan sensitivity determines whether a scan event is "countable" Possibly Suspicious Event Very Suspicious Event Sensitivity (from policy) Normal Event Low Count Medium Count Count High Count Count Count Countable scan events count against an origin source IP address Total number of countable events for all scan event types is compared to policy thresholds If threshold exceeded for a single IP address, policy-directed notification and documentation is triggered Scan instance event classification by event type included in appendix A Page 14

15 Attacks Against The TCP/IP Stack The system already silently defends itself from many attacks against the TCP/IP stack. IDS adds capability to control recording of intrusion events and to provide supporting documentation. IDS adds controls to detect and disable uncommon or unused features which could be used in an attack. Page 15

16 Attack Categories Malformed packet events Detects packets with incorrect or partial header information Inbound fragment restrictions Detects fragmentation in first 88 bytes of a datagram IP protocol restrictions Detects use of IP protocols you are not using that could be misused IP option restrictions Detects use of IP options you are not using that could be misused UDP perpetual echo Detects traffic between UDP applications that unconditionally respond to every datagram received ICMP redirect restrictions Detects receipt of ICMP redirect to modify routing tables. Outbound RAW socket restrictions Detects z/os RAW socket application crafting invalid outbound packets Flood Events Detects flood of SYN packets from "spoofed" sources Detects high percentage of packet discards on a physical interface Page 16

17 Attack Policy Overview Attack policy provides the ability to: Control attack detection for one or more attack categories independently Generate notification and documentation of attacks Notify the installation of a detected attack via console message or syslogd message Trace potential attack packets Generate attack statistics on time interval basis Normal or Exception Control defensive action when attack is detected Page 17

18 Interface Flood Detection Packet discard rate by physical interface is tracked to determine if there is a potential attack A high percentage of discarded packets on a physical interface may indicate the interface is under attack. Notification and traces provided when a possible interface flood condition is occurring (according to the discard threshold value). Provides information to help determine the potential cause of the interface flood Narrows flood condition to a local interface so you can Vary the interface offline This action not controlled with IDS policy Start tracing flood back to source Source MAC address of the "prior hop" (for OSA QDIO and LCS devices) Source IP address from the outer IPSec header if the packet had been received as IPsec tunnel mode. Source IP address could be a gateway or firewall Could allow source tracking closer to the source than "prior hop" Page 18

19 Interface Flood Detection Process Policy related to interface flood detection Specified on Attack Flood policy 2 actions attributes provided IfcFloodMinDiscard (default 1000) IfcFloodPercentage (default 10) For each interface, counts are kept for The number of inbound packets that arrived over the physical interface The number of these packets that are discarded When the specified number of discards (IfcFloodMinDiscard) is hit: If the discards occurred within one minute or less: the discard rate is calculated for the interval : # discards during the interval / # inbound packets for the interval If the discard rate equals or exceeds the specified threshold, an interface flood condition exists If discards occurred during period longer than 1 minute, not a flood condition Once an interface flood is detected, this data is collected and evaluated for the interface at 1 minute intervals. The interface flood is considered ended if the discards for a subsequent interval: Fall below the minimum discard value OR Discard rate for the interval is less than or equal to 1/2 of the specified threshold Page 19

20 Interface Flooding Example Assume the IDS flood policy specifies: IfcFloodMinDiscard: 2000 IfcFloodPercentage:10% Consider the following sequence for interface X: time time interval > 1 min inbound cnt discard cnt discard rate notes 13, N/A took longer than a minute to see the minimum discard count, so not a flood and discard rate not calculated. < 1 min 30, % not a flood, rate <10% < 1 min 20, % interface flood start detected. Run 1 minute timer until flood end detected. +1 min 40, % flood condition still exists, reset 1 minute timer. +1 min 50, % Interface flood end detected. Discard rate <= half of policy specified rate. Page 20

21 Traffic Regulation for TCP Allows control over number of inbound connections from a single host Can be specified for specific application ports Especially useful for forking applications Independent policies for multiple applications on the same port e.g. telnetd and TN3270 Connection limit expressed as Port limit for all connecting hosts AND Individual limit for a single connecting host Fair share algorithm Connection allowed if specified individual limit per single remote IP address does not exceed percent of available connections for the port All remote hosts are allowed at least one connection as long as port limit has not been exceeded QoS connection limit used as override for concentrator sources (web proxy server) Page 21

22 TCP connection regulation algorithm Configured maximum allowed connections for a given port: N In-use connections Available connections: A Connections from a given source IP address: IP@x: X Configured controlling percentage: CP If a new connection request is received and A=0, the request is rejected. If a new connection request is received and A>0 and the request is from a source that already has connections with this port number (in this example: IP@x), then: If X+1 < CP*A then Allow the new connection Else Deny the new connection Purpose: If close to the connection limit, then a given source IP address will be allowed a lower number of the in-use connections. Page 22

23 Regulation algorithm example Source IP address X attempts its fifth connection Allowed Rejected Total Allowed Connections Available CP=10% CP=20% CP=30% A B A B If we currently 40 connections available (A=40) and a controlling percentage (CP) of 20%, when source IP address X tries to establish its fifth connection, it will be allowed (40 * 20% = 8, so 5 connections is within the acceptable range). If we have 20 connections available (A) and CP is again 20%, when source IP address X tries to establish its fifth connection, it will be rejected (20 * 20% = 4, so 5 would exceed the allowable number of connections). Page 23

24 Traffic Regulation for UDP Allows control over length of inbound receive queues for UDP applications Specified on a per-port basis Can be applied to ports of your choosing Before TR for UDP, UDP queue limit control was requested globally for all queues UDPQueueLimit ON OFF in TCP/IP Profile If neither TR UDP or UDPQueueLimit is used, a stalled application or a flood against a single UDP port could consume all available buffer storage TR UDP supercedes UDPQueueLimit specification TR UDP queue limit expressed as abstract queue length VERY SHORT SHORT For applications that tend to receive data faster than they can process it LONG VERY LONG Useful for fast or high priority applications with bursty arrival rates Page 24

25 z/os Communications Server Security IDS Actions Recording actions Defensive actions Page 25

26 Recording Actions Recording options controlled by IDS policy action specification Options Event logging Syslogd Number of events per attack subtype recorded in a five minute interval can be limited Local Console Recording suppression provided if quantity of IDS console messages reach policy-specified thresholds Statistics Syslogd Normal and Exception conditions IDS packet trace Activated after attack detected Number of packets traced for multipacket events are limited Amount of data trace is configurable (header, full, byte count) All IDS events recorded in syslog and console messages, and packet trace records have probeid and correlator Probeid identifies the point at which the event detected Correlator allows association of corresponding syslog and packet trace records Page 26

27 Defensive Actions by Event Type Scan Events No defensive action defined Attack Events Packet discard Certain attack events always result in packet discard and are not controlled by IDS policy action malformed packets flood (synflood discard) Some attack types controlled by IDS policy action ICMP redirect restrictions IP option restrictions IP protocol restrictions IP fragment outbound raw restrictions perpetual echo No defensive action defined flood (interface flood detection) Traffic Regulation Events Controlled by IDS policy action TCP - Connection limiting UDP - Packet discard Page 27

28 IDS and Defensive Filtering The Defense Manager component allows authorized users to dynamically install time-limited, defensive filters: A local security administrator can install filters based on information received about a pending threat Enables filter installation through automation based on analysis of current attack conditions Defensive filtering is an extension to IDS capabilities Adds additional defensive actions to protect against attacks z/os Security Administrator DM defensive filter database z/os Defense Manager Initial filters installed via TCP/IP Profile and/or Policy Agent Enable dynamic defensive actions on z/os ipsec command Maintain defensive filters ipsec command TCP, UDP IP Applications Filter rules IDS IDS Network Interfaces Automation software Message processing Requires minimal IP Security configuration to enable IP packet filtering function Uses ipsec command to control and display defensive filters Defense Manager Manages installed defensive filters in the TCP/IP stack Maintains record of defensive filters on DASD for availability in case of DM restart or stack start/restart Defensive filter scope may be: Global - all stacks on the LPAR where DM runs Local - apply to a specific stack Defensive filter are installed "in-front" of configured/default filters Page 28

29 z/os Communications Server Security Intrusion Detection Reports for Analysis Page 29

30 IDS Log Reports trmdstat command produces reports based on IDS data recorded in syslog Types of reports generated for logged events Overall summary reports Connection and IDS Event type summary reports For Connection, Attack, Flood, Scan, TCP and UDP TR information Event type detail reports For Connection, Attack, Flood, Scan, TCP and UDP TR information Types of reports generated for statistics events Details reports Attack, Flood, TCP and UDP TR reports Page 30

31 Tivoli Support for IDS Events Tivoli NetView provides local z/os management support for IDS NetView provides ability to trap IDS messages from the system console or syslog and take predefined actions based on IDS event type such as: Route IDS messages to designated NetView consoles notifications to security admistrator Run trmdstat and attach output to Issue pre-defined comands Tivoli Security Information and Event Manager (TSIEM) provides enterprise-wide management support for IDS Automated aggregation and correlation of events, logs, and vulnerabilities Broad device support for multi-vendor environments, including security, network, host, and applications Support includes processing for z/os Communications Server syslog messages for IDS events Automates policy and regulatory compliance Policy and Regulatory based policy monitoring and reporting Page 31

32 z/os Communications Server Security Working with IDS Policy Controlling, displaying, and validating policy Defining IDS policy IDS policy configuration with Configuration Assistant for z/os Communications Server example Page 32

33 Controlling Active IDS Policy Configurable policy deletion controls in Policy Agent configuration file TcpImage statement FLUSH NOFLUSH {PURGE NOPURGE} FLUSH and NOFLUSH take effect at Policy Agent initialization FLUSH - specifies that any active policy should be deleted NOFLUSH - specifies that active policy should not be deleted PURGE and NOPURGE take effect at Policy Agent termination PURGE - specifies that any active policy should be deleted NOPURGE - specifies that active policy should not be deleted Refresh Policy At Interval (1800-second default) specified on TcpImage statement With MODIFY PAGENT command (REFRESH option) When Policy Agent configuration file (HFS only) is updated (refresh is automatic) Page 33

34 Displaying IDS Policy pasearch command Displays IDS policy read by Policy Agent netstat command Displays installed IDS policy in TCP/IP stack Displays statistics by policy category Tip: Restrict access to IDS policy displays using SAF SERVAUTH resources: EZB.PAGENT.sysname.tcpname.IDS EZB.NETSTAT.sysname.tcpname.IDS Page 34

35 Steps for Validating IDS Policy 1. Inspect configured IDS policy for correctness 2. Invoke PAGENT and TRMD 3. Issue PASEARCH and verify that the correct policy is installed 4. Keep policy in force for a trial period 5. Issue IDS netstat to view active IDS policy and statistics 6. Run TRMDSTAT reports to verify syslog messages for intrusion events 7. Adjust the policy as required Page 35

36 Defining IDS Policy GUI-based approach to configuring: IDS AT-TLS IPSec and IP filtering QoS Policy-based Routing (PBR) Defense Manager Daemon (V1R11) Focus on high level concepts vs. low level file syntax Runs on Windows and under z/osmf (V1R11) Builds and maintains Policy files Related configuration files JCL procedures and RACF directives ( V1R11) Supports import of existing policy files (V1R10) Download the Windows-based Configuration Assistant at: Page 36

37 IDS Policy Configuration Steps with the Configuration Assistant 1. Download and install the Configuration Assistant configuration tool 2. Configure IDS policies a. Examine IDS defaults and base policy on defaults b. Copy IDS defaults into a new IDS requirements map c. Make changes to new requirements map as needed 3. Create system image and TCP/IP stack image 4. Associate new requirements map with TCP/IP stack 5. Perform policy infrastructure and application setup tasks 6. Transfer IDS policy to z/os Page 37

38 Configuration Assistant for z/os Communications Server Page 38

39 Start a new IDS configuration Page 39

40 Create IDS policy objects Page 40

41 Evaluate IDS_Default requirements map IDS_Default provided as default requirement map Display details of the requirement map Evaluate whether they meet your requirements Page 41

42 Details view of IDS_Default requirements map (1 of 4) Page 42

43 Details view of IDS_Default requirements map (2 of 4) Page 43

44 Details view of IDS_Default requirements map (3 of 4) Page 44

45 Details view of IDS_Default requirements map (4 of 4) Page 45

46 Use IDS_Default as a starting point Using IDS_Default as a base Copy IDS_Default Create new requirements map using copied IDS_Default as a base Page 46

47 Name new requirements map Page 47

48 Modify copied default requirements map next page Page 48

49 Attack protection enabled by default next page Page 49

50 Customize report settings Page 50

51 Enable scan policy next page Page 51

52 Modify Global Scan Settings Page 52

53 Enable traffic regulation protection No traffic regulation defaults Policy selections are system dependant System capacity a consideration in setting maximum limits Page 53

54 Define TCP TR policy for FTP next page Page 54

55 Set details for TR Page 55

56 Traffic regulation enabled Page 56

57 IDS_Policy_Demo requirements map now created Page 57

58 Create System Image Page 58

59 Create TCP/IP stack Page 59

60 Associate TCP/IP Stack with Requirements Map next page Page 60

61 Perform application setup tasks Page 61

62 Install configuration files next page Page 62

63 Show the configuration file to be installed Page 63

64 Set up to transfer policy file to z/os Page 64

65 z/os Communications Server Security Features Summary Page 65

66 IDS Features Summary IDS events detected include: Scan detection TCP port scans UDP port scans ICMP scans Sensitivity levels for all scans can be adjusted to control number of false positives recorded. Attack detection Malformed packet events Outbound raw restrictions Inbound fragment restrictions IP option restrictions IP protocol restrictions ICMP redirect restrictions Flood events (Synflood, Interface flood) UDP perpetual echo Traffic Regulation UDP backlog management by port Packets discard TCP total connection and source percentage management by port Connection limiting IDS recording options Event logging syslogd, local console Statistics syslogd normal, exception IDS packet trace after attack detected for offline analysis Number of packets traced for multi-packet events are limited Reports and event handling trmdstat produces reports from IDS syslogd records Summary and detailed IDS event handling by Tivoli NetView Tivoli Security Information and Event Manager Defensive filtering (z/os V1R10) Installed through ipsec command Manually (by human being) or through automation (via external security event manager) Page 66

67 z/os Communications Server Security Appendix A Scan Probe Instance Event Classifications Page 67

68 ICMP Scan Probe Instance Classification Request Type Destination Address Event Classification any subnet base or broadcast very suspicious Information req single host possibly suspicious Subnet Mask req single host possibly suspicious Echo with IP Option Record Route single host possibly suspicious Echo with Record Timestamp Echo or Timestamp, denied by QOS policy single host single host possibly suspicious normal Echo or Timestamp single host normal Page 68

69 UDP Scan Probe Instance Classification Socket State Event Event Classification RESERVED to no one Unbound, not RESERVED recv any packet recv any packet very suspicious possibly suspicious - app may be temporarily down Bound packet rejected by QOS policy normal Bound packet rejected by FW filtering possibly suspicious Bound recv any packet normal Page 69

70 TCP Scan Probe Instance Classification Socket State Event Event Classification Any state recv unexpected flags (SYN+FIN...) very suspicious RESERVED recv any packet very suspicious Unbound, not RESERVED recv any packet possibly suspicious - app may be temporarily down Listen recv SYN classification deferred if syn queued. Half open connection recv ACK normal - connection handshake completed Half open connection Half open connection Any connected state Any connected state Any connected state recv RST possibly suspicious - scanner covering tracks? final time out (and not syn flood) seq# out of window recv standalone SYN final time-out very suspicious - scanner abandoning handshake? normal - perhaps duplicate packet normal - perhaps peer reboot possibly suspicious - peer abandoned connection Page 70

71 For More Information... URL Content IBM Communications Server Twitter Feed IBM Communications Server Facebook Fan Page IBM System z in general Networking: IBM zseries Servers IBM Software Communications Server products z/os Communications Server Communications Server for Linux on zseries Communication Controller for Linux on zseries Communications Server products - white papers, product documentation, etc ITSO redbooks IBM z/os Communications Server technical Support including TechNotes from service Technical support documentation from Washington Systems Center (techdocs, flashes, presentations, white papers, etc.) Requests For Comment (RFC) IBM z/os Internet library PDF files of all z/os manuals including Communications Server Page 71